Based on the high intubation rate reported above, it is important to know when NIV should not be applied in patients with acute hypoxemic respiratory failure. Most of the studies that reported the experience of using NIV in acute hypoxemic respiratory failure have proposed predictors of NIV failure. In 2001, Antonelli et al. investigated in a prospective multicenter cohort study factors involved in NIV failure [41]. In a heterogeneous population, the overall efficacy of NIV in avoiding intubation (70%) contrasted with the high rate of failure observed in 86 patients fulfilling the diagnosis of ARDS (51%). The intubation rate was similar among patients with ARDS of pulmonary versus extrapulmonary origin, but sepsis on admission was associated, among other variables, with NIV failure. In a large prospective observational study of NIV in 147 ARDS patients in experienced centers (NIV failure rate around 50%), a high Simplified Acute Physiology Score (SAPS) II and PaO₂/FiO₂ ≤ 175 mmHg 1 h after initiation of NIV were independently associated with NIV failure [42]. Rana et al. [38] found that NIV success was significantly correlated with low severity scores (Acute Physiology and Chronic Health Evaluation III or Sequential Organ Failure Assessment), with high PaO₂/FiO₂ (147 vs. 112), and with less pronounced acidosis than patients with NIV failure. Last, Thille et al. found that shock, active cancer, low level of consciousness, and mild/severe ARDS were independently associated with NIV failure [39].

The efficiency of NIV in patients with acute hypoxemic respiratory failure due to ALI, ARDS, or severe pulmonary infiltrates, and for whom endotracheal intubation is not mandatory, depends on the degree of hypoxia, the presence of comorbidities and complications, and the illness severity scores. It also probably depends on the respiratory drive of the patients. The high rate of NIV failure suggests a cautious approach with these patients, consisting of early NIV trial and no delay of needed intubation. Measurement of tidal volume under NIV may be important, and patients having tidal volume above 8 mL/kg of predicted body weight may be at higher risk of failure and of having an injurious breathing pattern. In those patients, lung-protective ventilation may be considered as a therapy. In addition, when using NIV in a patient with acute hypoxic respiratory failure, in an attempt to avoid intubation, one should always consider the risks of inappropriate intubation delay. Demoule et al. found that the effect of NIV differs between acute hypoxemic respiratory failure (including mainly in this definition, ALI and ARDS) and patients with cardiogenic pulmonary edema or acute exacerbation of COPD, because NIV failure was associated with higher mortality in patients with acute hypoxemic respiratory failure [32]. This finding should therefore raise a note of caution when applying NIV for this indication and make the clinician wonder whether a lung-protective ventilation is not more appropriate. Close monitoring is therefore crucial when using this technique as a first-line therapy in patients with ARDS. Delaying intubation may contribute to mortality [43]. NIV should not be considered primarily as an alternative to invasive ventilation.

The most conventional form of oxygen delivery relies on face masks, nasal cannula, or nasal prongs. However, some drawbacks limit their use in case of severe hypoxemia if oxygen flow higher than 15 L/min is needed and in case of high patients’ inspiratory flow that may induce a certain amount of oxygen dilution. An alternative to conventional oxygen therapy has received growing attention: heated, humidified high-flow nasal cannula oxygen (HFNC) is a technique that can deliver heated and humidified oxygen, with a controlled FiO₂, at a maximum flow of 60 L/min of gas via nasal prongs or cannula. Many data with this technique have been published in the neonatal field where it is increasingly used [44]. Since a decade, the use of HFNC has been considered for patients with acute hypoxemic respiratory failure. The high flow rates used generate low levels of positive pressure in the upper airways [45, 46]. The high flow rates may also decrease physiological dead space by flushing the expired carbon dioxide from the upper airway [47, 48], a process that potentially explains the observed decrease in the respiratory rate and the work of breathing [49]. In patients with acute respiratory failure of various origins, HFNC has been shown to result in better comfort and oxygenation than standard oxygen therapy delivered through a face mask [50–52]. HFNC has gained increasing clinical and scientific interest [36, 50–53]. The larger study on the use of HFNC in patients with acute hypoxemic respiratory failure was conducted by Frat and coworkers in 2015 [36]. In this randomized controlled trial, investigators aimed at determining whether HFNC administered through a large-bore close-fitting nasal cannula or NIV could reduce the intubation rate and improve outcomes in acute hypoxemic patients compared with standard oxygen administration [36]. In this trial, patients had mild or moderate ARDS (PaO₂/FiO₂ ratio around 155 mmHg) mainly due to pneumonia. The primary outcome, the rate of endotracheal intubation, was lower among patients treated with high-flow oxygen than among those who received standard oxygen therapy or NIV, but the rates did not differ significantly (38% vs. 47% and 50%, respectively) (p = 0.18). However, in a post hoc adjusted analysis that included the 238 patients with severe initial hypoxemia (PaO₂/FiO₂ ≤200 mmHg), the intubation rate was significantly lower among patients who received high-flow oxygen than among patients in the other two groups ( p = 0.009). Furthermore, the hazard ratio for death at 90 days after randomization was 2.01 in the standard oxygen group versus the high-flow oxygen group ( p = 0.046) and 2.5 in the NIV group versus the high-flow oxygen group ( p = 0.006). One explanation proposed by the authors was that NIV and spontaneous breathing could have provided tidal volume greater than 9 mL/kg of predicted body weight. Hence, the degree of lung injury might have been increased in this group, contributing to a higher mortality than that observed in the high-flow oxygen group.

Acute respiratory failure remains the most common and severe life-threatening complication in immunocompromised patients [54, 55]. Many immunocompromised patients with acute respiratory failure require ventilatory support within a few hours after admission to the ICU. Avoiding invasive mechanical ventilation significantly decreases the risk of death. Thus, choosing the optimal device for delivering oxygen is of the utmost importance. In a large multicenter study of 1,004 patients with solid or hematologic malignancies and ARDS meeting the new operational Berlin definition, Azoulay et al. reported that NIV was used initially in one-third of the patients. They frequently failed, with the highest failure rates occurring in the most severe ARDS category [55]. Nevertheless, a favorable impact of NIV has been described in retrospective studies [56–58] and in a few randomized controlled trials including a relatively small number of patients [25, 26, 59] (Table 15.2). Hence, the efficacy of NIV in this population appears to be promising. More recently, the use of HFNC has also been investigated in immunocompromised patients. Below, we will discuss the advantages and limits of using NIV and HFNC in immunocompromised patients.